Accepted Version downloaded from SOAS Research Online: http://eprints.soas.ac.uk/23000/
Tibetan *-as > - os
Nathan W. Hill1
Abstract: Both Jacques (2010) and Zeisler (2015) propose explanations for the synchronically unexpected past zos of the Tibetan verb 'eat'. After evaluating their proposals, this essay suggests that zos is the regular outcome of a sound change *as > - os, the results of which were erased through analogy in almost all other verbs.
Tibetan verbs showing stem ablaut typical have -o- in the present and -a- in the past, e.g. 'kill' with present gsod, past bsad, future gsad, and imperative sod.
The verb 'eat' with the stems za, zos, bzaḥ , zo has the opposite pattern showing -a- in the present and -o- in the past. Following Meillet's (1925: 25) principle that irregular morphology preserves archaisms, Guillaume Jacques (2010) proposes that zos is a fragment of erstwhile agreement morphology in Tibetan. His proposal has not proven popular. Randy LaPolla (2012: 120) objects to the importance that Jacques places on this one verb.2 LaPolla's objection is misplaced for two reasons. First, single verb forms are sometimes of paramount significance for an entire family; witness Vedic śáye 'lies' (Clackson 2007:
146). Second, 'eat' is not the only Tibetan verb to show this pattern. Hill (2014) draws attention to three further verbs that appear to show a vowel -o- in the past, viz. ḥ deṅ , doṅ , —, — 'disappear', ḥ chaḥ , ḥ chos, ḥ chaḥ , ḥ cho 'chew, gnaw', and laṅ , loṅ s, laṅ 'finish' (cf. Hill 2010: 89, 148, 279).
In addition to LaPolla, Zeisler (2015) also rejects Jacques' explanation of zos, instead arguing that zos is borrowed into the paradigm of 'eat' from the cognate potentialis verb 'be able to eat'. Of the additional verbs that Hill (2014) notes, Zeisler (2015: 43-44), points to the confused and contradictory reports of the indigenous lexicographical tradition for 'disappear' and 'finish' to speculate that Hill conflates separate verbs. Zeisler finds 'chew' “most interesting, particularly as it seems to display the same pattern as the verb za
‘eat’” but regrets that it is “not very well attested” (2015: 44). She notes two attestations of a present stem ḥ cha, both in the phrase rus-pa gle ḥ cha ‘the bones, fodder for the gle’ (Pt 1194, ll. 62-3 and IOL Tib J, r68).
Zeisler fails to consult the Wörterbuch der tibetischen Schriftsprache, which provides ample attestations of this verb. The Wörterbuch, a research project of the Bayerischen Akademie der Wissenschaften, started in 1954, began publication in 2005, and at the time of writing in July 2015 has grown to 24 fascicles,
1 I would like to thank Abel Zadoks for first proposing to me that Tibetan -a- changes to -o- in (some) closed syllables.
2 For his part, LaPolla offers no alternative explanation for zos.
published by peniope: http://www.peniope.de/ijdl.htm
Accepted Version downloaded from SOAS Research Online: http://eprints.soas.ac.uk/23000/
reaching bsñol. This dictionary supersedes all previous Tibetan lexicographical work in coverage and scientific rigor. To showcase the excellence of this
resource I quote the three relevant entries in extenso.
bcaḥ fut. zu ↓²ḥ chaḥ beißen, kauen, essen; ~ ba Nahrung, Getränk, Saufen; bzaḥ ~ Speisen und Getränk, Essen.
bzaḥ ~ daṅ ni na bzaḥ … gsol (metr.) „er gab Speisen und Getränk sowie Kleider“ (Anav 1: 89,15); bzaḥ ba daṅ ~ ba gya nom pa
„üppiges Fressen und Saufen“ (Prav 187,7); bzaḥ ba daṅ btuṅ ba daṅ ~ ba rnam pa sna tshogs „verschiedenartige Speisen, Getränke, und Nahrung (skt. bhojya)“ (Suv 95,21); bzaḥ ~ de bźin btuṅ ba ñid ..
rab tu bzaḥ (metr.) „Essen und ebenso auch Triken soll man zu sich nehmen“ (Hev 1.6.20a); so yi dag byed ~ bar bya (metr.) „mann soll [die Zweige] zum Reinigen der Zähne kauen“ (Ahs 1.2.3b); mgo bo na bas ḥ di mi ~ (metr.) „bei Erkrankungen des Kopfes soll man dies nich essen“ (Ahs 1.2.4b); bzaḥ ba daṅ / ~ ba … kyis yaṅ dag par tshim par byas nas „als er sie mit Speisen und Getränken völlig
zufriedengestellt hatte“ (ViśṬ 76,36).
Lex. bzaḥ ba daṅ ~ ba ma ḥ oṅ s par lhuṅ bzed mi bzed ≅ nānāgate khādanīye bojanīye pātram upanāmayiṣ yāmaḥ „solange die Zeit zu essen oder zu trinken nicht gekommen ist, werden wir nicht die Bettelschale hinhalten“ (Mvy 8569); gźib pa ni sos ~ ba lces ḥ jib pa „bźib pa:
mit den Zähnen beißen, mit der Zunge saugen“ (Kloṅ D 736,6).
²ḥ chaḥ fut. ↑²bcaḥ knabbern, kauen, essen; vgl. ↓²ḥ chos.
rtsa ba ḥ bras bu ~ ba daṅ (metr.) „Wurzeln und Früchte knabbern“
(Prav 51.21); rus la ~ baḥ i rus kyaṅ dkon (metr.) „Knochen sind selten, sogar für diejenigen, die sie essen“ (gZer 510,6).
Lex. ~ ≅ sos ldad pa sogs (Dagy).
²ḥ chos essen, kauen; vgl. ↑²ḥ chaḥ
~ sam zos na źes bya ba ni lkog mar kham gis mid naḥ o „gekaut oder gegessen bedeutet: man schluckt die Speise den Hals hinunter“ (K5 297a6).
Lex. khādita „gekaut“ (in Mvy 7040); ~ pa ≅ carvita (Ak 288.60); ~ pa
≅ myaṅ s paḥ am zos paḥ i don du ḥ aṅ snaṅ „wahrgenommen oder erscheint auch i. S. v. gegessen“ (brDa); ~ pa ≅ zos pa (TTC).
As the reader sees, the Wörterbuch falls short of labeling ḥ chos the past of ḥ chaḥ but the equation of ḥ chos-pa with zos-pa in the sources cited provides evidence that ḥ chos is a past stem.
A few passages from the Kanjur further confirm that ḥ chaḥ , ḥ chos, and bcaḥ belong to the same verb and are respectively present, past, and future.
dge-sloṅ -gis rnam-pa gsum dben-par bya-ste / bśaṅ -ba-daṅ / gci-ba-
Accepted Version downloaded from SOAS Research Online: http://eprints.soas.ac.uk/23000/
daṅ / so-śiṅ bcaḥ -baḥ o / dge-sloṅ -gis so-śiṅ ḥ chos-nas / de bźin-du mi dor-gyi chus bkru-bar bya /
A monk shall go into isolations for three purposes, defecation, urination, and chewing on toothpicks. After a monk has chewed a
toothpick, he is to wash with water which has not been thus discarded.
(K8, Vol. 13, 310a)
yaṅ dge-sloṅ -ma gaṅ dus ma yin-par bcaḥ -baḥ am bzaḥ -ba ḥ chaḥ - ḥ am za-na ltuṅ byed-do
Also, if a nun chews and eats what is to be chewed and eaten when the time is not full, this is a transgression. (K4, Vol. 9, 14b)
The Kanjur also attests the imperative, which the Wörterbuch omits.
tshe-daṅ ldan-pa ḥ di ḥ cho śig / ḥ di zo źig ces bya-ba ni bcaḥ -ba- daṅ bzaḥ -baḥ o/
When (someone) says, 'O venerable sir, chew this! Eat this!' the (offering) is to be chewed and to be eaten (K3, Vol. 7, 148a)
These citations from the Kanjur and the occurrence of bcaḥ and ḥ chos in the Mahāvyutpatti (Mvy), published before 814 CE, taken together with Zeisler's own citation of ḥ cha in Dunhuang documents, guarantee that the entire paradigm is of hoary provenance. The conjugation of 'chew' ḥ chaḥ , ḥ chos, bcaḥ , ḥ cho that these citations establish is parallel to 'eat' za, zos, bzaḥ, zo.
Zeisler's explanation that zos and ḥ chos are borrowed form a potentialis paradigm is possible, but poorly motivated. The postulated independent verbs *zo 'to be able to eat' and *ḥ cho 'be able to chew' are as far as I know
unattested. It is unclear why speakers would target the past stem and not some other form for this replacement by borrowing from a potentialis. It seems unlikely that an inherited bzas would yield to such a borrowing, since
analogical pressure (e.g. bsams, bsgrubs, etc.) reinforces it as the expected form. The distribution of zos in peripheral dialects versus bzas in the center (Zeisler 2015: 46) suggests that zos is the archaism. In addition, syllable structure weighs against Zeisler's proposal of borrowing the potentialis as a past stem. In an earlier paper, where she first draws attention to the
potentialis, Zeisler (2002) notes the potentialis verbs chod 'able to cut', sod 'able to kill', lon 'able to take', and sñogs 'able to catch'. These verbs are all closed syllable whereas the two verbs that show -o- ablaut in the past have open syllable roots. Perhaps this distribution is coincidence, but, perhaps not.
An account of this distribution is a plus for any explanation of the ablaut seen in 'eat' and 'chew' and Zeisler's explanation does not garner this plus.
Presuming that Jacques would see ḥ chos as additional evidence for erstwhile agreement, his explanation accounts for the attested phonological distribution.
He offers three concrete possibilities for the phonological development of the -
published by peniope: http://www.peniope.de/ijdl.htm
Accepted Version downloaded from SOAS Research Online: http://eprints.soas.ac.uk/23000/
os in zos; 1. *zaus > zos, 2. *zau > *-zo with -s added by analogy, 3. *zasu >
zos, presumably with an intermediate phase such as *zosu (2010: 47). None of these laws leads to -o- in closed syllable roots. Consider the verb √laṅ 'take' len, blaṅ s, blaṅ , loṅ s; the three proposals all produce the attested past: 1. *blaṅ us > blaṅ s, 2. *blaṅ u > *blaṅ → blaṅ s, 3. *blaṅ su >
blaṅ s. These proposals require one to consider all transitive past stems ending in -as as analogical developments. Thus, in the verb 'do' byed, byas, bya, byos, the innovative past byas replaced inherited *byos, which was lost without a trace. A verb such as 'think' sems, bsams, bsam, soms serves as an analogical model for byas, viz. bsam : bya :: bsams : X = byas.
Each of Jacques' proposals has ramifications for Tibetan historical phonology in general. The first proposal requires a sound change *-us > -s, that operated after *-au- > -o-, to avoid *zaus developing to *zas instead of zos. This sound change leaves unexplained why some words still contain -us, such as rus 'bones'.
The second and third proposal require the lost of final -u, a not implausible change per se, but one which gives rise to the problem that words such as bu 'son', su 'who', and ḥ bru 'grain' did not undergo the change. Zeisler
reasonably objects to the third proposal that if *asu becomes -os then one might expect *isu to develop to *esu or another outcome other than the -is seen in verbs such as 'do' bgyid, bgyis, bgyi, gyis (2015: 46). Each of Jacques's
proposals is rather complicated and partly unmotivated in its details. Ockham's razor favors abandoning the insistence on a *-u- suffix and accepting a simpler sound change, namely *as > -os.
The proof of a phonological account for the forms zos and ḥ chos is whether the account explains idiosyncrasies other than those that served as its
motivation. A change *-as > -os has the advantage of explaining the invariant verb ltos 'look to, attend to' as the inherited past of lta, bltas, blta, ltos 'look at'.3 An original paradigm lta, ltos, blta, ltos closely parallels za, zos, bzaḥ , zo and the innovative past bltas parallels the innovative past bzas.
Postulating an inherited paradigm zlo, *zlas > zlos, bzla, zlos similarly reconciles the two verbs zlo, bzlas, bzla, zlos 'say, repeat' and zlos
(invariant) 'repeat'. The pair of verbs dgaḥ 'be happy' and dgos 'need, want', both of invariant conjugation, suggests an intransitive verb with the
conjugation dgaḥ , dgos, lacking a distinct future and imperative as non-
volitional verbs do. In a more complicated case, for the verb smra, smras, smra, smros 'say' we predict an inherited past stem *smros. Although *smros does not exist as a separate verb, there is an invariant verb smos 'say, call' which the Dag yig gsar sgrigs sees as additionally an alternative present (sic) of smra.
One may legitimately speculate that smos is a regular phonetic development from
3 The ensuing discussion proceeds with the hypothesis *-as > -os, but most of the argument still holds mutatis mutandis using Jacques' more complicated phonological proposals.
Accepted Version downloaded from SOAS Research Online: http://eprints.soas.ac.uk/23000/
*smras.4
Just as Jacques' proposals must either account for words ending in -u or -us, so too the proposal *as > -os must account for all instances of -as in the language. Analogy within a paradigm explains verb forms ending in -as,5 but cannot explain the case markers or nouns that have this rime. The case markers - las and -nas pose no particular problem. They are derived by the the suffixing of -s to the case markers -la and -na (Simon 1941: 385). This suffixation occurred after the change *as > -os. Nouns that end in -as arose after the application of this sound change, whether through borrowing or through
derivation. For example, a nominalizing -s forms the noun ltas 'omen' from the verbal root √lta 'see'; compare skyems 'beer, libation' from √skyem 'be thirsty' (Beyer 1992: 118). The supposition that *-as > -os is an old change answers the objection that ltas 'omen' does not relate to 'look' transparently enough to suggest a recent formation. Similarly zas 'food' derives via
suffixation from √za 'eat'. Such nouns as skas 'stairs', las 'deed', nas
'barely', sṅ as 'pillow', and ras 'cotton' lack recognized cognates elsewhere in the family.6 Tibetan ḥ bras 'rice', deriving from *ḥ mras according to Simon's law (Hill 2011: 448-449), has a Chinese cognate 糲 ljejH < *[m]ə-rˤ ats (21-26g) 'rice'. In this case, Tibetan final -s likely originates from the *-ts cluster seen in Chinese.7 If this explanation is correct, then final *-ts simplified to -s only after the change *-as > -os. The anteriority of *-as > -os to *ts > -s provides further support for an early operation the former. The Kurtöp cognates bù 'do', ɲù 'borrow', zù 'eat', chú 'devour' the generalized past forms cognate to Tibetan *byos (replaced by byas), *rños (replaced by brñas), zos, and ḥ chos further support an early date for the change *as > -os since it must have
occurred prior to the split of Tibetan and the East Bodish languages (Hyslop 2011: 55-56, 1247, 143).8 In sum, the comparative evidence poses little obstacle to, and potentially supports, the proposed change *-as > -os.
The inherited paradigms proposed here, together with brief remarks on
4 The relationship between smra, smos, and the additional verbum dicendi with the stems rma, rmas requires further attention.
5 There are a few verbs which synchronically speaking have a root final -s (mkhas 'know' (v.), glas 'change one's residence', ḥ gas, bkas, dgas, khos 'split (vt.)', ḥ gas, gas 'split (vi.)', ḥ gras 'feel revulsion, be unhappy', brñas 'ridicule, belittle', ḥ thas 'hard, firm', gdas 'speak', gnas 'stay (v.), place (n.)', spras, spras, spras, spros 'adorn, decorate', bas-pa 'finished, complete', gzas 'prepare to, be about to'), but the possibility remains that this -s was originally a past tense suffix in these conjugations.
6 The words in rime -as in Zhang (1985) not yet mentioned here are: klas-pa 'boundless, huge', gras 'class, type', gźas 'song', śas 'some, sharecropper field', slas 'retinue'.
7 Sagart (2014) points to a similar correspondence in the pair Tibetan rus, 'bone' and Chinese 律 lwit < *[r]ut (31-18c) 'pitch pipe', with the complication that Chinese is missing final
*-s.
8 Gong's (2002[1995]: 115) proposal to relate Tibetan rdzas 'thing, object' to Chinese 事 dzriH
< *[m-s-]rəʔ-s (0971a) 'serve; service, affair' is neither semantically nor phonetically compelling.
published by peniope: http://www.peniope.de/ijdl.htm
Accepted Version downloaded from SOAS Research Online: http://eprints.soas.ac.uk/23000/
subsequent developments, are as follows:
√za 'eat' pres. za
past. zos (exists alongside analogical bzas) fut. bzaḥ
imp. zo
√ḥ cha 'chew' pres. ḥ chaḥ past. ḥ chos fut. bcaḥ imp. ḥ cho
√lta 'look at' pres. lta
past. ltos (continues as separate invariant verb 'look at, attend to', replaced by analogical bltas)
fut. blta imp. ltos
√zla 'say, repeat' pres. zla
past. zlos (continues as separate invariant verb 'repeat', replaced by analogical bzlas)
fut. bzla imp. zlos
√dga 'be happy'
pres. dgaḥ (continues as separate invariant verb 'be happy') past. dgos (continues as separate invariant verb 'need, want')
√smra 'say' pres. smra
past. smos (continues as separate invariant verb 'say, call', replaced by analogical smros)
fut. smra
imp. *smos (obsolete, replaced by analogical smras)
√bya 'do' pres. byed
past. *byos (replaced by analogical byas, but compare Kurtöp cognate bù 'do')
fut. bya imp. byos
Accepted Version downloaded from SOAS Research Online: http://eprints.soas.ac.uk/23000/
√rña 'borrow' pres. rña
past. *rños (replaced by analogical brñas, but compare Kurtöp cognate ɲù 'borrow')
fut. brña imp. rños
Abbreviations
Ahs = Vogel (1965)
Ak = Vidyābhuṣ aṇ a (1911) Anav = Hofiinger (1982-1990)
brDa = Dge bśes chos kyi grags pa (1957) Dagy = Dagyab (1966)
Dag yig gsar sgrigs = Tsan chung (1979) gZer = Tenzin Namdak (1965)
Hev = Snellgrove (1959) K = Derge Kanjur,
http://thlib.org/encyclopedias/literary/canons/kt/catalog.php#cat=d/k Kloṅ D = Chandra (1973)
Mvy = Ishihama and Fukuda (1989), but following the numbering of Sakaki (1916) Prav = Eimer (1983)
Suv = Nobel (1944) TTC = Zhang (1985)
ViśṬ = Schneider (1993: 74-270)
References
Beyer, Stephen (1992). The Classical Tibetan language. New York: State University of New York.
Chandra, Lokesh, ed. (1973). “Rig gnas cha ba sgra rig pa / sñan ṅ ag / sdabs sbyar / zlos gar / mṅ on brjod / brdaḥ gsar rñiṅ gi khyad par rnams las byuṅ baḥ i miṅ gi graṅ s.” The Collected works of Longdol Lama. New Delhi : International Academy of Indian culture. ba, 1-16.
Clackson, James (2007). Indo-European Linguistics: An Introduction. Cambridge:
Cambridge University Press.
Dagyab, Loden Sherab (1966). Bod brdaḥ i tshig mdzod. Dharamsala: Dagyab.
Dge bśes chos kyi grags pa (1957). Brda dag miṅ tshig gsal ba. Beijing:
Nationalities Publishing House.
Eimer, Helmut, ed. (1983). Rab tu 'byuṅ ba'i gži: die tibetische Übersetzung des Pravrajyāvastu im Vinaya der Mūlasarvāstivādins. Wiesbaden: Otto Harrassowitz.
Gong Hwang-cherng. 2002[1995]. "The System of Finals in Proto-Sino-Tibetan". The Ancestry of the Chinese Language. William S. Y. Wang, ed. (Journal of
published by peniope: http://www.peniope.de/ijdl.htm
Accepted Version downloaded from SOAS Research Online: http://eprints.soas.ac.uk/23000/
Chinese linguistics. Monograph series 8) Berkeley: Project on Linguistic Analysis, University of California: 41-92. (reprinted in:)
漢藏語硏究論文集 Collected Papers on Sino-Tibetan Linguistics. Taipei:
中央硏究院語言學硏究所籌備處, 79-124.
Hill, Nathan W. (2010). A Lexicon of Tibetan Verb Stems as Reported by the Grammatical Tradition. Munich: Bayerische Akademie der Wissenschaften.
Hill, Nathan W. (2011). “An Inventory of Tibetan Sound Laws.” Journal of the Royal Asiatic Society of Great Britain & Ireland (Third Series) 21.4: 441- 457.
Hill, Nathan W. (2014). “Sino-Tibetan: Part 2 Tibetan.” Lieber, Rochelle and Štekauer, Pavol, (eds.), The Oxford Handbook of Derivational Morphology.
Oxford: Oxford University Press. 620-630.
Hofinger, Marcel, ed. (1982-1990). Le congrès du lac Anavatapta. Vies de saints bouddhiques. Extraits du Vinaya des Mūlasarvāstivādin, Bhaiṣ ajyavastu. 2 tomes. Louvain-la-Neuve: Université Catholique de Louvain, Institut
Orientaliste.
Hyslop, Gwendolyn (2011). A grammar of Kurtöp. PhD dissertation. University of Oregon.
Ishihama Yumiko and Fukuda Yōichi, ed. (1989). A New critical edition of the Mahāvyutpatti: Sanskrit-Tibetan-Mongolian dictionary of Buddhist
terminology. Tokyo: Toyo Bunko.
Jacques, Guillaume (2010). “A possible trace of verb agreement in Tibetan”.
Himalayan Linguistics 9.1: 41-49.
LaPolla, Randy J. (2012). “Comments on methodology and evidence in Sino-Tibetan comparative linguistics.” Language and Linguistics 13.1: 117-132.
Meillet, Antoine (1925). La méthode comparative en linguistique historique.
Oslo: H. Aschehoug & co.
Nobel, Johannes, ed. (1944). Suvarnaprabhāsottamasūtra. Das Goldglanz-Sūtra: ein Sanskrittext des Mahāyāna-Buddhismus. Band 1: Tibetische Übersetzung.
Stuttgart: Kohlhammer.
Sagart, Laurent (2014). “A Note on Tibeto-Burman Bone Words and Chinese Pitch- pipes.” Studies in Chinese and Sino-Tibetan Linguistics: Dialect,
Phonology, Transcription and Text. Richard VanNess Simmons and Newell Ann Van Auken, eds. Taipei: Institute of Linguistics, Academia Sinica, Taipei, Taiwan. 179-183
Sakaki Ryōzaburō 榊亮三郎 (1916). 梵藏漢和四譯對校飜譯名義大集 Bon-Zō-Kan-Wa yon'yaku taikō hon'yaku myōgi taishū. Kyoto: 眞言宗京都大學 Shingonshū Kyōto Daigaku.
Schneider, Johannes, ed. (1993). Der Lobpreis der Vorzüglichkeit des Buddha.
Bonn: Indica et Tibetica Verlag.
Simon, Walter (1941). “Certain Tibetan suffixes and their combinations.”
Harvard Journal of Asiatic Studies 5: 372-391.
Snellgrove, David L. (1959). The Hevajra Tantra: a critical study. London:
Oxford University Press.
Accepted Version downloaded from SOAS Research Online: http://eprints.soas.ac.uk/23000/
Tenzin Namdak, ed. (1965). Ḥ dus pa rin po cheḥ i rgyud gzer mig. New Delhi.
Tenzin Namdak.
Tsan chung (1979). Dag yig gsar sgrigs. Xining: mtsho sṅ on mi rigs dpe skrun khaṅ .
Vogel, Claus, ed. (1965). Vāgbhaṭ a's Aṣ ṭ āṅ gahṛ dayasaṃ hitā; the first five chapters of its Tibetan version. Wiesbaden: F. Steiner.
Vidyābhūṣ aṇ a, Satis Chandra, ed. (1911). Amarakoṣ a and its Tibetan translation: 'Chi med mdzod. Calcutta: Asiatic society.
Zeisler, Bettina (2002). “The development of temporal coding in Tibetan: some suggestions for a functional internal reconstruction. (1): Unexpected use of the 'imperative' stem in Old Tibetan and Themchen (Amdo Tibetan).”
Henk Blezer, ed., Tibet, Past and Present. PIATS 2000: Tibetan Studies:
Proceedings of the Ninth Seminar of the International Association for Tibetan Studies. Leiden: Brill, 441-453.
Zeisler, Bettina (2015). “Eat and drink – if you can! A language internal explanation for the ‘irregular’ paradigm of Tibetan za, zos, zo
‘eat’.” Himalayan Linguistics 14.1: 34-62.
Zhang Yisun (1985). Bod rgya tshig mdzod chen mo. Beijing: Minzu chubanshe.